hbertv2-Massive-intent

This model is a fine-tuned version of gokuls/bert_12_layer_model_v2_complete_training_new on the massive dataset. It achieves the following results on the evaluation set:

• Loss: 0.9457
• Accuracy: 0.8515

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 64
• eval_batch_size: 64
• seed: 33
• distributed_type: multi-GPU
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.1277	1.0	180	1.0263	0.7364
0.9042	2.0	360	0.8013	0.7875
0.6379	3.0	540	0.8182	0.7914
0.4865	4.0	720	0.8074	0.7973
0.3637	5.0	900	0.7780	0.8190
0.3019	6.0	1080	0.7656	0.8288
0.2218	7.0	1260	0.8253	0.8254
0.1741	8.0	1440	0.8295	0.8239
0.1316	9.0	1620	0.8590	0.8308
0.1011	10.0	1800	0.8465	0.8431
0.078	11.0	1980	0.9007	0.8401
0.0573	12.0	2160	0.9133	0.8470
0.0382	13.0	2340	0.9233	0.8470
0.0247	14.0	2520	0.9365	0.8490
0.0148	15.0	2700	0.9457	0.8515

Framework versions

• Transformers 4.30.2
• Pytorch 1.14.0a0+410ce96
• Datasets 2.13.0
• Tokenizers 0.13.3